# Mapping algorithm; availability depends on BWA version
"Algorithm":
StringIn(("backtrack", "mem", "bwasw"), default="backtrack"),
# Minimum mapping quality (PHREAD) of reads to retain
"MinQuality":
IsUnsignedInt(default=0),
# Remove unmapped reads or not
"FilterUnmappedReads":
IsBoolean(default=True),
# Use seed region during mapping
# Verbose name for command-line option "-l 65535"
"UseSeed":
IsBoolean(default=True),
# Any number of user specific options
StringStartsWith("-"):
Or(IsListOf(IsStr, IsInt, IsFloat),
Or(IsStr, IsInt, IsFloat, IsNone)),
},
"Bowtie2": {
# Minimum mapping quality (PHREAD) of reads to retain
"MinQuality":
IsUnsignedInt(default=0),
# Remove unmapped reads or not
"FilterUnmappedReads":
IsBoolean(default=True),
# Any number of user specific options
StringStartsWith("-"):
Or(IsListOf(IsStr, IsInt, IsFloat),
Or(IsStr, IsInt, IsFloat, IsNone)),
},
def test_string_starts_with__default_set__must_meet_spec():
with pytest.raises(ValueError):
StringStartsWith("FooBar", default="BarFoo")
msa = mkfile["MultipleSequenceAlignment"]
defaults = msa.pop("Defaults")
defaults.setdefault("Program", "MAFFT")
defaults["MAFFT"].setdefault("Algorithm", "MAFFT")
for key in mkfile["Project"]["Regions"]:
msa[key] = fill_dict(msa.get(key, {}), defaults)
unknown_regions = set(msa) - set(mkfile["Project"]["Regions"])
if unknown_regions:
raise MakefileError("Unknown Regions of Interest in Genotyping: %s" \
% (", ".join(unknown_regions),))
# Recursive definition of sample tree
_VALIDATION_SUBSAMPLE_KEY = And(StringStartsWith("<"),
StringEndsWith(">"))
_VALIDATION_SAMPLES_KEY = And(IsStr, Not(_VALIDATION_SUBSAMPLE_KEY))
_VALIDATION_SAMPLES = {
_VALIDATION_SAMPLES_KEY: {
"GenotypingMethod": StringIn(("reference sequence",
"random sampling",
"samtools"),
default="samtools"),
"SpeciesName": IsStr, # Not used; left for backwards compatibility
"CommonName": IsStr, # Not used; left for backwards compatibility
"Sex": IsStr(),
"Gender": IsStr(),
}
}
_VALIDATION_SAMPLES[_VALIDATION_SUBSAMPLE_KEY] = _VALIDATION_SAMPLES
def test_string_starts_with__default_not_set():
spec = StringStartsWith("Foo")
assert spec.default is DEFAULT_NOT_SET
def test_string_starts_with__default_set__valid_value():
spec = StringStartsWith("Foo", default="FooBar")
assert spec.default == "FooBar"
def test_string_starts_with__rejects_not_uppercase_str(value):
spec = StringStartsWith("Foo")
with pytest.raises(MakefileError,
match="Expected value: a string with prefix 'Foo'"):
spec(_DUMMY_PATH, value)
def test_string_starts_with__rejects_string_without_prefix():
spec = StringStartsWith("A_")
with pytest.raises(MakefileError):
spec(_DUMMY_PATH, "B_GHI")
def test_string_starts_with__accepts_standard_str():
spec = StringStartsWith("A_")
spec(_DUMMY_PATH, "A_BC")
_PATH_IN_EXCEPTION_VALUES = (
(IsInt(), "foo"),
(IsUnsignedInt(), -1),
(IsFloat(), "abc"),
(IsBoolean(), 1),
(IsStr(), 1),
(IsNone(), 1),
(ValueIn([1]), 2),
(ValuesIntersect([1]), [2]),
(ValuesSubsetOf([1]), [2]),
(ValueMissing(), True),
(And(IsStr), 1),
(Or(IsStr), 1),
(Not(IsInt), 1),
(StringIn("abc"), 1),
(StringStartsWith("FOO"), 1),
(StringEndsWith("FOO"), 1),
(IsListOf(IsInt), "foo"),
(IsDictOf(IsInt, IsInt), 1),
)
@pytest.mark.parametrize("spec, value", _PATH_IN_EXCEPTION_VALUES)
def test_specs__path_is_displayed_in_exception(spec, value):
with pytest.raises(MakefileError, match=_DUMMY_PATH_STR):
spec(_DUMMY_PATH, value)
###############################################################################
###############################################################################
# process_makefile
msa = mkfile["MultipleSequenceAlignment"]
defaults = msa.pop("Defaults")
defaults.setdefault("Program", "MAFFT")
defaults["MAFFT"].setdefault("Algorithm", "MAFFT")
for key in mkfile["Project"]["Regions"]:
msa[key] = fill_dict(msa.get(key, {}), defaults)
unknown_regions = set(msa) - set(mkfile["Project"]["Regions"])
if unknown_regions:
raise MakefileError("Unknown Regions of Interest in Genotyping: %s" %
(", ".join(unknown_regions), ))
# Recursive definition of sample tree
_VALIDATION_SUBSAMPLE_KEY = And(StringStartsWith("<"), StringEndsWith(">"))
_VALIDATION_SAMPLES_KEY = And(IsStr, Not(_VALIDATION_SUBSAMPLE_KEY))
_VALIDATION_SAMPLES = {
_VALIDATION_SAMPLES_KEY: {
"GenotypingMethod": RemovedOption(),
"SpeciesName": RemovedOption(),
"CommonName": RemovedOption(),
"Sex": IsStr(),
"Gender": RemovedOption(),
}
}
_VALIDATION_SAMPLES[_VALIDATION_SUBSAMPLE_KEY] = _VALIDATION_SAMPLES
# Genotyping settings; note that explicit lists must not be used here, to allow
# proper inheritance of default values. Use IsListOf instead.
_VALIDATION_GENOTYPES = {